them to us", and consequently that the genuinely antique "is very often preserved in its purest form in late, and even in the latest, authorities". It is to be regretted that Roscher has not formally stated on what principle he decides whether the version of a myth presented by a late authority is or is not primeval ; for though he sometimes acts on the right principle, he also sometimes acts on the wrong one. Indeed, it seems as though he had never faced this question of mythological method, or even realised the existence of the question. The result is that his book is valuable principally as a comprehensive collection of material from the classics. The value of his conclusions seems to me extremely uneven ; but let me place some of them before the reader, in order that he may judge for himself And I will begin by giving some instances in which Roscher has, as it seems to me, acted on the right principle — the principle that a belief, tale, rite, or custom, however late the authority for its existence, may be regarded as primitive, provided that it can be shown to exist, or have existed, among some other savage or primitive people. The wrong principle, which I will also illustrate from Roscher, is that a belief, etc., may be regarded as primeval because it appears "simple", or because it is common in classical poetry, or because Aristotle or Galen adopted it.
Thus, the belief that the spirits of the dead take up their abode in the moon, even if the belief were mentioned as existing in Greece by no authority earlier than Plutarch, would be rightly regarded, as it is regarded by Roscher, as primitive, on the ground that some South American tribes also entertain it. Readers of Folk-Lore, remembering Mr. Frazer's demonstration (vol. i, pp. 148 ff.) that the wisdom of Pythagoras was but the folk-lore of the peasant, will be pleased to find that Pythagoras also regarded the moon as the abode of the departed (Iamblichus, Vita Pythag., xviii, 82). Again, that the sun and moon are a pair of lovers, or a married couple, is a conception which is rightly vin-
